Closed geo-lit closed 1 year ago
I don't expect widespread usage of ECDAA. I think the underlying cryptography is relatively young, i.e. the 2021 draft fixes security holes from 2018. At the moment, I recomment to solve attestation related privacy issues by simply not using batch attestation. This is the default in OpenSK. It might fail for some websites with strict policies, but I don't expect OpenSK to be on their whitelists anyway. I'd argue that ECDAA's additional complexity is not worth the marginal benefit. Without attestation already, security keys are a better second factor than most.
That is for our product decision on ECDAA on OpenSK. However, from a research perspective, I welcome anyone to implement and experiment with it. A few thoughts and pointers:
process_make_credential
here. It should give you some ideas about how to replace it with a different mechanism.Feel free to tell me more about where exactly you're lost or ask more concrete questions, then I can give you a better answer where to look.
There are no new developments around ECDAA, so I assume no interest. Please reopen when this changes.
Hi I was in contact with @jmichelp asking him about ECDAA.
Just wondering if anyone knows why the draft has not been implemented, and has been stagnant since 2018 (although I recently found a new draft from FIDO from May 2021 https://fidoalliance.org/specs/common-specs/fido-ecdaa-algorithm-v2.1-rd-20210525.pdf) I know that it's not in the specification, but seems like it would solve any privacy issues that come with batch attestation, which is not 100% private.
Also looking on guidance in terms of how I could start it's implementation using OpenSK. I tried familiarizing myself with the repo, but am a bit lost, so any guidance would be appreciated, specifically attestation flow, and where ECDAA would be added from the program-flow perspective.